Prophylactic levofloxacin to prevent infections in newly diagnosed symptomatic myeloma: the TEAMM RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 08/116/69. The contractual start date was in September 2011. The draft report began editorial review in June 2017 and was accepted for publication in March 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Mark T Drayson reports personal fees from Abingdon Health (Abingdon Health, York, UK) outside the submitted work. Stella Bowcock reports personal fees from Amgen (Amgen, CA, USA) and Celgene (Celgene, NJ, USA) outside the submitted work, non-financial support to attend educational meetings and has a patent issued for a device broadly related to the work. Tim Planche reports personal fees from Pfizer (Pfizer, CT, USA), Actellion (Actellion, Allschnil, Switzerland) and Astellas (Astellas Pharma, Tokyo, Japan) outside the submitted work. Kwee Yong reports grants from Janssen (Janssen Pharmaceutica, Beerse, Belgium), Celgene and Chugai (Chugai Pharmaceutical Co Tokyo, Japan) outside the submitted work. David Meads is a member of the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) European Economic and Social Committee (EESC) Methods Group and the NIHR HTA EESC Panel. Claire T Hulme is a member of the NIHR HTA Commissioning Board. Janet A Dunn is a member of the NIHR Efficacy and Mechanism Evaluation board.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Drayson et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Background

Myeloma is a cancer of bone marrow plasma cells that causes anaemia, skeletal fractures, renal failure and profound immunodeficiency. There are approximately 5500 new cases of myeloma in the UK per annum. 1 However, the overall prevalence is increasing, given the improved survival rates over the past four decades. 2 The median age at presentation is approximately 70 years and only 15% of patients are aged < 60 years. Myeloma has a higher incidence in African-Caribbean ethnic groups than in Caucasians, but there are few other distinctive epidemiological features. 3 The majority of cases present de novo, but it is now recognised that this is preceded by an asymptomatic monoclonal gammopathy of undetermined significance phase in virtually all patients. 4

Myeloma causes profound immunodeficiency and recurrent serious infections. One-quarter of patients will have a serious infection within 3 months of diagnosis. Ten per cent of patients die within the first 60 days of diagnosis, with bacterial infection directly causing 45% of these deaths. 5 Recent advances in antimyeloma treatment have improved overall survival significantly, yet this high early-death rate remains little changed, affecting all prognostic groups. Patients who may have survived long term with current antimyeloma treatment are dying soon after diagnosis, with the biggest single cause being bacterial infection. Therefore, newly diagnosed myeloma patients may benefit from antibacterial prophylaxis to prevent infection, hospital admission and early death. Reducing infection may also improve response to antimyeloma treatment by reducing interruptions of antimyeloma treatment and reducing immune responses to infection that promote myeloma cell survival and growth. In patients with other causes of immunodeficiency, such as neutropenia, asplenia, human immunodeficiency virus infection or reflux nephropathy, the importance of prophylactic antibiotics to prevent infection is well established and the administration of prophyactic antibiotics is common practice in the NHS. However, their usefulness in myeloma has not been established. Furthermore, some of the studies that established the use of antibacterial prophylaxis in other conditions predate the current rise in health care-associated infections (HCAIs), such as Clostridium difficile. The data from these older trials may not reflect current risks associated with antibiotic prophylaxis and so there is a need to reassess the effect of antibiotic prophylaxis on HCAI.

Existing research

Large studies in Europe and North America have identified a high mortality rate (8–20%) in the first 3 months following a diagnosis of myeloma, with bacterial infection being the single biggest identifiable cause. 5–8 Analysis of 3107 myeloma patients registered into UK Medical Research Council (MRC) trials from 1980 to 2002 showed that 10% of patients died within 60 days of trial entry and that 45% of these deaths were directly attributable to bacterial infection. 5

In the ‘MRC myeloma 9’ trial,9 which recruited between 2003 and 2008, overall incidence of infection in non-intensively treated patients was 214 out of 692 (30.9%), with a median time to infection from first diagnosis of myeloma of 43 days. Recent advances in antimyeloma treatment have improved survival significantly, yet this high early-death rate has remained unchanged for > 30 years and affects all prognostic groups. This suggests that current supportive care strategies, including the treatment of an infection once established, may be insufficient. Streptococcus pneumoniae, Staphylococcus aureus and Escherichia coli are the most frequent types of bacterial infection in myeloma patients. 10–15 The risk of these infections is associated with myeloma disease activity and abates as the disease is brought under control with antimyeloma treatment.

The mechanism by which the risk of infection is increased in the presence of active myeloma disease is not well understood. Over 90% of 3218 MRC myeloma trial patients had reduced levels of normal antibodies,16 and these patients’ susceptibility to bacterial chest infections is characteristic of antibody deficiency. However, a previous MRC trial (MacLennan ICM, Chapman C, Hazelwood M, North J. University of Birmingham, 1993) of immunoglobulin G (IgG) replacement treatment (double-blind, randomised, placebo-controlled trial of 203 patients) did not significantly reduce mortality or morbidity from infection in the first 3 months after diagnosis, despite effectively increasing total serum IgG levels and titres against specific bacterial pathogens. Myeloma patients are not usually neutropenic at presentation, and, in one study, only 11 out of 135 myeloma patients dying of infection within 60 days of diagnosis had a neutrophil count of < 2.0 × 10⁹/l. 5 Other factors associated with active myeloma disease that might increase the risk of infection include low serum complement component 4 (C4) levels, increased transforming growth factor beta and increased interleukin 10. 17

Antibacterial prophylaxis is an obvious strategy to prevent infection, hospital admission and early death in these patients. Of the only two trials18,19 of prophylactic antibiotics in early myeloma, one prospective randomised study18 was with co-trimoxazole in the early 1990s. This showed a reduction in bacterial infections with prophylactic co-trimoxazole (2/28 treated vs. 11/26 control patients) but the sample size was too small to detect reduced mortality. A recent trial19 of 212 patients given ciprofloxacin, co-trimoxazole and placebo found no difference in the rate of infection. This study19 was, again, underpowered to show differences in infection and mortality. The low incidence of all infections (22%) in this study raises the question whether or not the patients were representative of the normal myeloma clinic population. A retrospective analysis20 of infections in 202 patients on new therapies found that 40% of patients had an infection within 6 months, with 80% of severe infections (16% of patients) occurring in the first cycle of treatment. Antibiotic prophylaxis was effective in preventing infections in those patients with surrogate markers of high tumour burden (monoclonal band of > 3 g/dl, platelet count of < 130 × 10⁹/l), but not in those without these parameters.

Antibiotic prophylaxis should be active against the bacteria commonly causing infections in the patients treated, should be given as ideally oral, once-daily medication to maximise adherence and efficacy, and should have few side effects. For all of the above reasons, the quinolones, particularly ciprofloxacin and levofloxacin, are now the most commonly used antibiotics for chemoprophylaxis.

Although less than one-tenth of myeloma patients dying of infection are neutropenic, the immunosuppressed state in both neutropenic and early myeloma patients leads to bacterial infection. 5 The common organisms causing infection in myeloma are E. coli, S. pneumoniae, Klebsiella spp., S. aureus, Pseudomonas spp., Haemophilus spp. and Proteus spp. These are similar to those organisms seen in neutropenic infections, although Gram-negative infections are more common in neutropenia. Thus, studies on the use of prophylactic antibiotics active against the common pathogens that cause infection in neutropenia are pertinent to myeloma patients.

A large meta-analysis21 including 162 studies with 12,599 neutropenic patients found that all antibiotic prophylaxis significantly reduced the risk of death compared with placebo or no treatment [relative risk (RR) 0.66, 95% confidence interval (CI) 0.55 to 0.79]. Fluoroquinolone prophylaxis was the most effective and reduced the risk of all-cause mortality (RR 0.52, 95% CI 0.37 to 0.74), as well as of infection-related mortality, fever, clinically documented infection and microbiologically documented infections. Fluoroquinolone prophylaxis increased the risk of adverse events (AEs) (RR 1.52, 95% CI 0.79 to 2.92), but these were minor events. The benefit of reduction in infection-related mortality (RR 0.49, 95% CI 0.31 to 0.77) far outweighed any mortality from adverse effects because all-cause mortality was still markedly reduced (RR 0.52, 95% CI 0.37 to 0.74). These studies translate into a number needed to treat of 50 (95% CI 34 to 268) in order to prevent one death from all causes in neutropenic patients.

To date, only two studies22,23 have reported differences in costs, and both showed a cost benefit for prophylaxis. These studies focused on individual resource use elements, such as the total cost of antibiotics or hospital inpatient days. None of the trials included a comprehensive cost analysis or a full economic evaluation.

Levofloxacin prophylaxis may, in addition to preventing infection, improve response to antimyeloma treatment. Delivery of antimyeloma treatment is often delayed by infection and so reducing infectious episodes may increase the amount of antimyeloma treatment given. There is epidemiological and laboratory evidence that the cytokines and inflammatory mediators associated with bacterial infection may promote the growth of myeloma cells. 17 By reducing infections, antibiotic prophylaxis may reduce myeloma growth and potentiate response to antimyeloma treatment. This will be the first trial to assess these factors.

However, quinolones, along with other antibiotics, are implicated in increased risk of colonisation with antibiotic-resistant bacteria and invasive infection by those bacteria. These HCAIs have been an ever-increasing problem to the NHS over the past 10 years, accounting for significant morbidity and mortality. Up to one in four people carry S. aureus, and C. difficile may be carried by 1–3% of healthy people. 24 Up to 30% of long-term hospitalised patients may carry C. difficile. There were 36,095 cases of C. difficile-associated diarrhoea in the UK in 2008–9. 24

There is an increasing perception that antibiotic prophylaxis will increase numbers of HCAIs. A Midlands survey (carried out by the TEAMM trial management group) found that 24 haematologists did not use antibiotic prophylaxis alongside conventional myeloma chemotherapy, whereas eight haematologists did so in selected patients (unpublished audit). Half of the haematologists routinely used antibiotic prophylaxis in patients receiving intensive myeloma chemotherapy. Guidelines for the diagnosis and management of multiple myeloma published in 2009 by the UK Myeloma Forum3 on behalf of the British Committee for Standards in Haematology state that:

. . . there is insufficient evidence to recommend the routine use of prophylactic antibiotics (Grade C recommendation; level IV evidence).

Bird et al. 3

There are insufficient data on the relationship between changes in carriage rate of potentially pathogenic organisms during antibiotic treatment and the risk of subsequent infection with the same organism. From meta-analysis25 on antibiotic prophylaxis trials in neutropenia, there was no significant increase in C. difficile infection (7/1250 patients receiving a fluoroquinolone prophylaxis vs. 5/1279 receiving placebo or no treatment). Furthermore, recruitment to these trials predates Tackling Early Morbidity and Mortality in Myeloma (TEAMM) and the current problems with HCAIs by > 7 years. Although recent European guidelines26 recommend fluoroquinolone prophylaxis in severe neutropenia, adherence to this recommendation is not universal. In trials in which resistance data have been reported, patients on fluoroquinolones did not develop more infections with pathogens resistant to the drug than patients on placebo (RR 1.04, 95% CI 0.73 to 1.5). By reducing the number of clinical infections, levofloxacin may reduce the total amount of antibiotics used in these patients and lessen the emergence of resistance. 22 Although the emergence of bacteria resistant to fluoroquinolones can occur in units using fluoroquinolone antibiotic prophylaxis, there are no clear data on whether or not patients are harmed as a result. 27,28

In summary, the above data show that fluoroquinolone prophylaxis in neutropenia is very effective, but there are concerns about inducing fluoroquinolone-resistant organisms and HCAIs. This supports the equipoise position for this trial. No substantial trial of antibiotic prophylaxis in myeloma has been undertaken. The proven efficacy of levofloxacin in neutropenic patients and the sensitivity to levofloxacin of bacteria that cause infection in myeloma indicate that levofloxacin prophylaxis will also be effective in myeloma. The higher absolute risk of early death in myeloma (≈10% in the first 12 weeks from diagnosis in some risk groups) suggests that antibiotic prophylaxis may be even more effective in myeloma than in neutropenia. As there is a need for such an antibiotic trial in myeloma, it provides an excellent opportunity to collect data on HCAIs and quantify absolute risk of colonisation and infection during antibiotic prophylaxis. Data from the proposed trial will help to inform rational decisions about risks and benefits of antibiotic prophylaxis in many areas of medicine.

Research objectives

To assess the risks, benefits and cost-effectiveness of levofloxacin in newly diagnosed symptomatic myeloma by a prospective, multicentre, randomised, double-blind, placebo-controlled trial.

Trial design

Tackling Early Morbidity and Mortality in Myeloma was a randomised, double-blind, placebo-controlled, multicentre phase III trial assessing the benefit of antibiotic prophylaxis and its effect on HCAIs in patients with symptomatic multiple myeloma. Target recruitment was originally 800 patients; however, this was extended to up to 1000 patients because of a high rate of recruitment and availability of investigational medicinal product (IMP). Patients were randomised in a 1 : 1 ratio to 500 mg of levofloxacin for 12 weeks or placebo to match. The primary outcome was time to first febrile episode or death. Secondary outcomes included, but were not limited to, response to antimyeloma treatment, carriage and invasive infections with C. difficile, MRSA and ESBL coliforms, QoL and overall survival.

The TEAMM intervention

Patient randomisation and blinding

Written informed consent was obtained for all patients recruited to the trial. Randomisation of participants occurred through the WCTU Randomisation Service. Treatment allocation was performed using a minimisation algorithm and was stratified by hospital, estimated glomerular filtration rate (eGFR) and intention to give high-dose chemotherapy with autologous stem cell transplant (Figure 1).

FIGURE 1.

Trial schema.

At the point of randomisation, eligibility of the patient was confirmed and a check was performed on the consenting doctor to make sure they had been appropriately trained and delegated by the centre’s PI. Confirmation of the patient’s participation in the trial was sent to their general practitioner (GP) by the research team at the site using an approved template letter provided by the trial team. Trial treatment allocation was blinded to the clinicians, patients and the co-ordinating centre. The trial statistician allocated the term ‘active’ or ‘placebo’ in a 50 : 50 split to a list of 1500 randomly generated drug pack numbers. This list was sent (with password protection) to the drug packaging company that put the active drugs or placebo in the correctly labelled packs. This list was then used to build the bespoke randomisation and drug inventory system after the terms ‘active’ and ‘placebo’ were changed to ‘A’ or ‘B’ for concealment purposes. The trial statistician retained the master list, which revealed the identity of ‘A’ and ‘B’. Ordering of the drug for each site was done via the database that picked at random an even number of ‘A’ pack numbers and ‘B’ pack numbers from those at the storage facility. Once these pack numbers were received by the sites, they were activated and available to be allocated to patients when randomised.

Emergency unblinding could be requested on grounds of safety by any clinician who was involved in the medical care of the patient. Emergency unblinding was performed by telephone contact with the Emergency Scientific and Medical Services team at Guy’s and St Thomas’ Hospital, which held a master unblinding list with the allocated pack numbers. Emergency unblinding was available 24 hours a day, 365 days a year, and was considered an option only when the patient’s future treatment required knowledge of the trial treatment assignment.

Treatment

All patients entering TEAMM received antimyeloma treatment and supportive care including bisphosphonates as per standard practice at their hospital trust. If it was intended that the patient would proceed to high-dose chemotherapy with autologous stem cell transplant, this information was collected at randomisation and taken into account during stratification.

When patients were within 14 days either side of starting a programme of antimyeloma treatment, they received 500 mg of levofloxacin or placebo-to-match tablets daily for 12 weeks (84 days). The start of the antimyeloma treatment was determined as the start of steroids or chemotherapy, whichever came first. In a situation in which patients’ renal function was compromised, it was recommended that patients took a reduced dose of the trial drug, as levofloxacin is eliminated from the body mainly via excretion of unmetabolised drug in the urine by the kidneys. eGFR, provided locally, when possible, was assessed at baseline and reassessed at each scheduled trial visit to identify changes in renal function that would necessitate a change in dose of levofloxacin. It was recommended that eGFR was assessed within the 7- to 14-day period prior to randomisation. Those patients with an eGFR of > 50 ml/minute/1.73 m² took two tablets once per day (a dose of 500 mg), those patients with an eGFR of 20–50 ml/minute/1.73 m² took one tablet daily (a dose of 250 mg) and those patients with an eGFR of < 20 ml/minute/1.73 m² took a half-tablet daily (a dose of 125 mg). The active and placebo tablets were in identical breakable form. Dose reductions were recorded on the front of the patient diary, which was provided at each trial visit in conjunction with a review of eGFR.

Patients were asked to complete diaries during and after the 12-week treatment period, which were used to capture information related to drug compliance, health resource use and febrile episodes. Patients were asked to take their temperature daily (at a similar time each day) using a digital oral thermometer provided by the co-ordinating centre. They were also asked to take and record their temperature at any time they felt unwell. If a temperature of ≥ 38 °C was recorded, they were encouraged to contact the hospital for assessment, whether or not antibiotic treatment was required. In the event of a febrile episode, it was suggested that patients remain on the trial drug and that management of infections should be as for an individual who was taking active levofloxacin. Patients were treated as per standard practice depending on the nature of the infection. On resolution of the infection, it was recommended that patients continue taking the trial drug. Any patient who had stopped taking the trial drug while being treated for an infection was asked to restart promptly on resolution of the infection. Any treatment breaks were recorded on the case report forms (CRFs) and excess tablets remaining as a result were returned to pharmacies for accountability purposes.

Supportive treatment

Supportive treatment practices common to each centre were allowed; this included the use of bisphosphonates, prophylactic antivirals and prophylactic Pneumocystis treatment. The use of prophylactic Pneumocystis treatment was discouraged and the use of nebulised pentamidine over oral co-trimoxazole (Septrin^®, Actavis, Barnstaple, UK) was preferred. Other antibacterial prophylaxis was not allowed.

Central laboratory assessments

Microbiology assessments

Microbiological analysis was conducted at the Department for Medical Microbiology at St George’s Hospital, London, and the Birmingham Public Health Microbiology (PMH) laboratory. Nasal swabs and stool samples were requested from patients at baseline (before the first dose of trial medication) and at 4, 8 and 12 weeks when receiving TEAMM treatment and again at 16 weeks. These were used to assess carriage of S. aureus, C. difficile and ESBL coliforms. Any toxigenic strains of C. difficile were identified by culture and ribotyping. Extended multilocus VNTR analysis typing will be performed on all isolates in the Birmingham PMH laboratory. ESBL-positive Gram-negative bacteria from faecal screens and clinical specimens (when available) were identified and sent to the Birmingham PMH laboratory for genotyping of CTX-M beta-lactamase genes using denaturing high-performance liquid chromatography. Nasal swabs were cultured for MRSA, and isolates were typed and stored. Cultures from invasive infections isolated locally were transferred to the central laboratory at St George’s Hospital for typing when possible.

Immunology assessments

Immunology analysis was conducted by the Clinical Immunology Service at the University of Birmingham. Ethylenediaminetetraacetic acid (EDTA) blood, clotted blood and urine were requested from patients at baseline (before the first dose of trial medication) and at 4, 8 and 12 weeks when receiving TEAMM treatment and again at 16 weeks and 1 year. These samples were used to assess paraprotein levels, prognostic factors and markers of immunocompetence.

Measurements at entry and at 8 and 12 weeks included levels of:

• whole and free light chain (flc) paraprotein in serum and urine

• β₂-microglobulin, albumin, creatinine, calcium and C-reactive protein (CRP)

• complement components C3 and C4

• acute-phase response proteins and cytokines

• serum levels of polyclonal immunoglobulin (specific antibody against panels of both bacterial and viral antigenic targets and type I natural antibody levels)

• single-platform flow cytometric enumeration of lymphocyte subsets including type I and type 2 B cells, memory B cells; gamma/delta, cluster of differentiation 4 (CD4) and CD8 T cells; naive and memory subsets; Treg cells; and C cells

• monocyte subsets defined by CD14 and CD16 dendritic cells

• buffy coat cells, plasma and serum aliquoted and stored at –80 °C (not measured at 8 weeks).

Measurements at 4 and 16 weeks included levels of:

• whole and flc paraprotein in serum and urine

• β₂-microglobulin, albumin, creatinine, calcium, CRP response, markers of inflammation, and humoral and cellular immunocompetence.

The schedule of assessments is shown in Table 1.

TABLE 1 - Schedule of investigation

ECOG, Eastern Cooperative Oncology Group; EORTC QLQ-C30, European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30; EORTC QLQ-MY24, European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-24-item myeloma-specific module; EQ-5D, EuroQoL-5 Dimensions; HADS, Hospital Anxiety and Depression Scale.

The timing of viisits was flexible to allow them to coincide with chemotherapy visits as far as possible.

Event	Start of trial treatment	Trial visits (post start of trial treatment)
		4 weeks (± 2 weeksa)	8 weeks (± 2 weeksa)	12 weeks (end of treatment)	16 weeks	12 months
Informed consent taken	✗
Medical history to include ECOG performance status and weight and comorbidities	✗	✗	✗	✗	✗	✗
Inclusion criteria satisfied	✗
Levofloxacin/placebo supplied to patient	✗
QoL (EQ-5D, EORTC QLQ-C30 and HADS)		✗	✗		✗
QoL (EQ-5D, EORTC QLQ-C30, EORTC QLQ-MY24 and HADS)	✗			✗
Patient diary supplied to patient (includes questions on health resource use)	✗	✗	✗
Post-treatment patient diary supplied to patient				✗
Compliance with trial medication assessed (counting of empty blister packs)		✗	✗	✗
Details of infections and hospital admissions collected	✗	✗	✗	✗
AEs		✗	✗	✗	✗
Details of supportive care collected	✗	✗	✗	✗	✗
12–20 ml of clotted peripheral blood, 8 ml of EDTA blood (at start of treatment), 4 ml of EDTA blood thereafter and 20 ml of urine to the University of Birmingham	✗	✗	✗	✗	✗	✗
Stool sample and nasal swab to St George’s Hospital	✗	✗	✗	✗	✗
Bone marrow aspirate ± trephine	✗
Full blood count	✗	✗	✗	✗	✗	✗
Biochemistry screen	✗	✗	✗	✗	✗	✗
eGFR using modification of diet in renal disease formula	✗	✗	✗	✗	✗	✗

Patient follow-up

Patients were followed up at approximately 4-week intervals during their 12 weeks of TEAMM treatment. They were seen again at 16 weeks and had a further follow-up at 1 year post TEAMM treatment start date.

Serious adverse events

Investigators were required to inform the WCTU about the occurrence of a SAE within 24 hours of becoming aware of the event. SAEs had to be reported if they occurred between the first dose of trial medication and up to 30 days after the last dose was taken. An AE was considered to be a SAE if one of the following conditions applied:

• results in death

• is immediately life-threatening

• requires hospitalisation or prolongation of existing hospitalisation

• leads to the development of any grade 4 non-haematological toxicity (excluding alopecia)

• results in persistent or significant disability or incapacity

• is otherwise medically significant (e.g. important medical events that may not be immediately life-threatening nor result in death or hospitalisation, but may jeopardise the patient or may require intervention to prevent one of the other outcomes listed above, excluding new cancers or result of overdose).

A serious adverse reaction (SAR) was defined as a SAE that has a definite, probable or possible causal relationship to levofloxacin. Causality was assessed by both a clinician at site and the chief investigator. SARs that were unexpected according to the summary of product characteristics for levofloxacin were considered to be suspected unexpected serious adverse reactions. For every AE symptom, an AE term and grade was applied by the site using the Common Terminology Criteria for Adverse Events (CTCAE) version 4.03.

The following SAEs were not required to be reported as they relate to myeloma and its treatment:

• disease progression

• disease-related deaths

• routine treatment or monitoring if the studied indication is not associated with any deterioration in condition

• treatment, which was elective or preplanned, for a pre-existing condition, not associated with any deterioration in condition

• general care not associated with any deterioration in condition

• treatment on an emergency outpatient basis for an event not fulfilling any of the definitions of ‘serious’ (as provided above) and not resulting in hospital admission

• hospitalisation for palliative care

• grade 4 haematological toxicity is an expected consequence of effective treatment and is required to be reported only if it fulfils the criteria of a SAE (as defined above)

• treatment (including hospitalisation or extension of hospitalisation) for transfusions or pain relief

• surgical interventions for skeletal-related events (e.g. fixation of fractures or vertebroplasty)

• skeletal-related events, including bone fractures, spinal cord compression and increased bone pain

• hypercalcaemia

• extravasation

• toxicities that meet serious criteria that developed prior to entry to the trial.

Patients who died on treatment or within 30 days of the last dose of treatment were reviewed separately by the chief investigator and by a consultant haematologist who was independent of the trial.

Patient withdrawal

It was made clear in the patient information sheet (PIS) that patients were free to withdraw at any point after consenting to take part without having to give a reason, and that withdrawal would not affect the standard of care they would receive. When the withdrawal reason was known, it was supplied to the trial office via the withdrawal form. Three different withdrawal options were given to maximise data collection and options were discussed with the patient when possible:

1. Patient withdrew from treatment only – this option meant that the patient continued on the trial with all active assessments other than administration of the trial drug.

2. Patient withdrew from treatment and assessments – this option meant that the patient did not complete any assessments or take an active role in the trial, but was happy for us to collect data about their disease status and standard treatment they were receiving. This will enable long-term follow-up to be conducted on these patients.

3. Patient withdrew consent – for these patients, no further data were collected or used in the analysis after the date of withdrawal. Data collected between the date of consent and the date of consent withdrawal have still been used unless specified by the patient.

Patients could also have been withdrawn from the trial at the discretion of the chief investigator and/or Trial Steering Committee if safety was a concern.

Patients moving out of area

When patients moved area and were no longer attending visits at the same centre, every effort was made to transfer the follow-up of patients to the new centre where their treatment was continuing. This was possible only if the new centre had the relevant approvals to participate in TEAMM.

Outcomes

Sample size

The primary and first set of secondary outcomes were reached within 12 weeks of trial treatment. The primary outcome measure was time to first febrile episode or death from all causes, using a Kaplan–Meier survival curve. Assuming that the proportion of patients having a febrile episode or death was 30% in the first 3 months and that prophylactic antibacterials would reduce that rate to 20%, recruiting 800 patients into the trial (400 in each arm) would allow differences in excess of 10% to be detected with 90% power using a two-sided test at the 5% level of significance. Recruiting 1000 patients into the trial (500 in each arm) would allow differences in excess of 8% to be detected with 90% power using a two-sided test at the 5% level of significance. Recruiting 1000 patients would also allow detection of a levofloxacin-induced threefold increase in the rate of C. difficile-positive stools (from 5% to 15%) from entry to the trial to 12 weeks, with 95% power and 5% level of significance (two-sided test).

Other analyses include incidence of probable infections with site, severity and treatment; response to antimyeloma treatment and its relationship to infection; patient characteristics and indices of immunocompetence (blood leucocyte subset enumeration and antibacterial antibody titres) as prognostic markers for colonisation and invasive infection by antibiotic-resistant organisms; health economics; and QoL (by daily diary card and 4-weekly EQ-5D up to 16 weeks). With 1000 patients, it would be possible to report reliable estimates for these secondary outcomes.

Statistical methods

The main analysis, comparing time to first febrile episode or death from all causes within the first 12 weeks, was carried out using a log-rank comparison, with the start time being the date on which the patient started trial treatment to the time of a reported event, or to a censor date for those with no events reported after 12 weeks. 34 All randomised patients were included in the analysis of the primary end point as ITT using the date of randomisation as the start date for any patients not starting trial treatment. 35

The secondary end points such as C. difficile-containing stools, MRSA and ESBL coliform carriage rates and number of invasive infections associated with the identical organism previously carried were assessed using chi-squared tests with continuity adjustments. Mantel–Haenszel tests for combining 2 × 2 tables were then used to adjust for stratification variables and various prognostic factors. Patients who were randomised and had started treatment were included in the analyses of the secondary end points.

Overall survival was calculated from the date on which the patient started trial treatment to the date of death or date of censorship, as appropriate. Overall survival analysis was based on all-cause mortality and assessed using Kaplan–Meier curves. 34 The main treatment effect was assessed using the log-rank test. Kaplan–Meier curves for the primary outcome were constructed for each treatment arm and Cox proportional hazards models were used to compare arms after adjustment for stratification variables and imbalances in important baseline factors. 34,36 Covariates were assessed graphically for non-proportionality of hazards. When there was an indication of non-proportionality, a time-dependent variable (its product with time) was introduced to test non-proportionality. If the Wald chi-squared p-value was < 0.05 for this term, then non-proportionality was assumed and a restricted mean survival time approach used with the test of the difference in restricted means presented instead of the hazard ratio (HR). 37,38 These analyses were carried out on an ITT basis.

The analyses of all other secondary end points, incidence of probable infections with site, severity and treatment, response to antimyeloma therapy and its relationship to infection and indices of immunocompetence (blood leucocyte subset enumeration and antibacterial antibody titres) were undertaken using the appropriate statistical analyses tools.

Statistical analyses were performed using SAS^® version 9.4 (SAS Institute Inc., Cary, NC, USA). SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ^® indicates USA registration.

Independent Data and Safety Monitoring Committee

An independent Data and Safety Monitoring Committee (DSMC) was established, consisting of an independent statistician, haematologist and microbiologist. Their main objective was to advise the Trial Steering Committee if there is evidence or reason why the study should be amended or terminated based on recruitment rates, compliance, safety or efficacy. The DSMC met after the first 50 patients were recruited, and annually thereafter. Confidential reports containing recruitment, protocol compliance, safety data and interim analyses of outcomes (not formally tested outside the trial statistical analyses plan, which was agreed with the DSMC) were reviewed by the DSMC. Interim analyses of the primary outcome were presented to the DSMC using conservative tests with significance determine by a p-value of 0.001 (to preserve the overall alpha level of 0.05). All analyses were blinded to the trial statistician and the DSMC until agreement to unblind at the end of the trial.

The original power calculations aimed to detect a difference of 10% (i.e. from 30% in the control arm to 20% in the treatment arm) with 90% power at the two-sided 5% level of significance. The first planned look at the primary end point when 150 patients had completed the 12-week assessment indicated that the original assumptions held true (i.e. a 31% rate in the control arm vs. a 21% rate in the treatment arm; documented in the DSMC December 2013 report). The prespecified look at the data before the trial recruitment closed (i.e. 760 patients recruited and 642 recruited patients having completed the 12-week assessment) indicated that the rate in the placebo arm may be reduced to 23% (note that this analysis remained blind to the treatment arm). After much discussion, the DSMC made recommendations to increase the sample size from 800 patients to up to 1000 patients (the maximum that could be accommodated by the available drug supply) within the current funding window. Increasing the sample size from the original 800 patients to up to 1000 patients allowed the power to be retained at 80% with the ability to detect differences in excess of 7% depending on the final rate in the control arm.

Database and data processing

The database was held on WCTU’s Microsoft SQL Server (Microsoft Corporation, Redmond, WA, USA) system and imposed rules for data entry, which included having a valid range for responses, linked dates and patient identification (ID) numbers.

Data were single entered into the database by trial personnel. Checks were carried out on 100% of data entered by new starters and, following full training, 10% of data entered by trial personnel were checked each month. Unacceptable error levels in 10% checks were followed up with further checking and retraining. The trial statistician carried out checks of plausibility of values and missing data to enable further queries to be resolved prior to freezing data for scheduled analyses.

Screening and recruitment

Recruitment

A total of 977 patients were recruited between 3 August 2012 and 29 April 2016 from 93 NHS Hospitals (levofloxacin, n = 489; placebo, n = 488). Figure 2 shows actual compared with target recruitment.

FIGURE 2.

Cumulative recruitment.

Two patients who were randomised withdrew consent on the day of randomisation and were withdrawn from further analyses as no data were collected. Twenty-five patients did not start trial treatment; baseline information was collected for 23 of these patients, and these patients are included in the analysis of the primary end point under ITT.

Consolidated Standards of Reporting Trials flow diagram

The Consolidated Standards of Reporting Trials (CONSORT) flow diagram is shown in Figure 3.

FIGURE 3.

The CONSORT flow diagram.

Baseline characteristics

Protocol deviations/non-compliance

One patient was retrospectively found not to be eligible because they were found to have multiple plasmacytoma and not myeloma when further results came to light. A total of 24 patients were, at some point during their treatment, taking the incorrect dose on the basis of their eGFR result. Among this group, clinician decisions not to increase the dose based on patients’ conditions are classed as deviations. The remainder of instances of non-compliance were down to site errors or patient confusion. Details of the type of non-compliances are shown in Table 8.

Protocol deviations are shown in Table 9. A number of patients did not take the TEAMM medication for the correct number of days as stated in the protocol (84 days), some took it for fewer days than expected and some took it for more days than was stated. Those who took TEAMM treatment for fewer days than expected were required to note this either in a withdrawal form or on the on-treatment form as missed treatment. Patients who took the trial medication for more than the 84 days of treatment misunderstood the trial procedure and continued to take remaining tablets as a result of treatment breaks or dose reductions. Patients taking trial treatment for up to 98 days were not classed as non-compliers as it was felt that the data would not be significantly affected by this. The remaining patients who took trial treatment for > 98 days were classed as protocol violations. There were 13 patients who failed to begin their randomised treatment within the 14-day eligibility window stated in the protocol. All of these patients were eligible at the time of randomisation and all violations resulted from a change in the patient’s circumstances that resulted in the anticipated start date changing to one outside the 14-day window. These non-compliances are balanced by treatment arm, as shown in Table 9.

Treatment data across 12 weeks

Data were captured on participants while on treatment at 4, 8 and 12 weeks. At the time of data lock, a total of 2599 treatment forms had been returned by 925 patients. Table 10 presents on-treatment information by treatment arm.

The treatment information for the entire treatment period (from 0 to 12 weeks) is summarised in Table 11.

Infections

Deaths and infection-related deaths

Overall, a total of 116 deaths were reported among the 977 randomised patients (levofloxacin, n = 61; placebo, n = 55). A total of 30 patients (total of 52 causes) died within 12 weeks of starting trial treatment (levofloxacin, n = 8; placebo, n = 22). One patient randomised to placebo died prior to starting trial treatment. A total of 86 patients died post 12 weeks (levofloxacin, n = 53; placebo, n = 33). Table 20 presents causes of death by treatment arm. A total of 220 causes of death were reported for the 116 deaths, as selection of multiple causes is permitted on the death form. Cause of death was not available for one patient on placebo who died in a nursing home.

The majority of deaths were attributed to myeloma-related causes (n = 143, 65%) as opposed to non-myeloma-related causes (n = 77, 35%). Infection-related cause of death was less common in the levofloxacin arm (three causes) than in the the placebo arm (eight causes) within 12 weeks. However, more infection-related causes of death were reported in the levofloxacin arm (19 causes) than in the placebo arm (12 causes) post 12 weeks.

All deaths within 12 weeks were reviewed by an independent clinician to ascertain causes of death. Deaths post 12 weeks were reviewed by the two clinical chief investigators.

Time to first event by treatment arm across 12 weeks

Table 21 shows the number of febrile episodes, deaths and febrile episodes and deaths combined by treatment arm within the first 12 weeks. A total of 210 patients reported febrile infections (levofloxacin, n = 91; placebo, n = 119) out of the total 977 patients randomised.

Time to first event was calculated from the date of starting trial treatment to the date of febrile episode or date of death for the 229 patients reporting a febrile episode or death within 12 weeks. The remaining patients were censored at date of withdrawal or the last date considered appropriate. Figure 4 shows the 19% of patients randomised to levofloxacin reporting a febrile episode, or death, within 12 weeks from the start of trial treatment compared with the 27% in the placebo arm; log-rank χ² test = 9.78; HR 0.66, 95% CI 0.51 to 0.86; p = 0.002 in favour of levofloxacin.

FIGURE 4.

Time to first event (febrile or death) within 12 weeks. Log-rank χ² test = 9.78; HR 0.66, 95% CI 0.51 to 0.86; p = 0.002.

The Cox regression analysis showed that treatment is the most important factor and still retains significance in a multivariate model adjusting for baseline factors. A Cox regression model excluding treatment indicates that no other factors predict time to febrile episode or death. Assessing the individual factors indicates that ECOG performance status is the only factor with a borderline significance. Table 22 shows the treatment effect unadjusted and adjusted by baseline factors. It can be seen that adjustment makes little difference to the overall treatment effect.

The most significant individual baseline factor was the use of prophylactic Septrin (HR 0.59, 95% CI 0.44 to 0.80; p = 0.0008), followed by treatment (HR 0.66, 95% CI 0.51 to 0.86; p = 0.002). Adjusting treatment for prophylactic Septrin use made little difference to the treatment effect, indicating that these two variables have independent prognostic value (Figure 5).

FIGURE 5.

Treatment adjusted for Septrin (p = 0.002). (a) Septrin; and (b) no Septrin.

Overall survival

Overall survival was calculated from the date of starting trial treatment to the date of death or being censored at date last alive. Figure 6 shows that 98% of patients on levofloxacin survived for 12 weeks, compared with 95% of patients on placebo (p = 0.94). Twelve-month survival was similar across arms (levofloxacin, 90%; placebo, 91%). Follow-up of surviving patients was the same [median 12 months; interquartile range (IQR) 11–13 months] in both arms.

FIGURE 6.

Overall survival, by treatment. Log-rank χ² test = 0.004; p = 0.94.

Micro-organism sample return and acquisition

Table 23 shows the numbers of C. difficile, MRSA and ESBL Gram-negative organisms that were present at baseline. It also shows the number of new acquisitions between baseline and 16 weeks. The total number of nasal and stool samples returned between baseline and 16 weeks is indicated in the left-hand column. There were no differences in new ESBL acquisitions between the placebo and levofloxacin arms (30 vs. 25, respectively). There were no differences in acquisitions for C. difficile (8 vs. 11) and MRSA (7 vs. 4) between the placebo and levofloxacin arms.

Days on additional anti-infective treatment and total doses taken for treatment of infection, by treatment arm

A summation of all anti-infective drugs prescribed for the treatment of infections during the trial treatment period by treatment arm, excluding TEAMM trial medication, is shown in Table 24.

Response to antimyeloma treatment

Response to myeloma treatment within the first 12 weeks is an early indication of whether or not patients will reach stable disease. The main reason why patients do not respond is disease progression. Changes between blood samples collected at baseline and the last sample collected within 12 weeks will indicate if patients have had an early response to antimyeloma treatment. This will be presented as part of the TEAMM National Institute for Health Research (NIHR) Efficacy and Mechanism Evaluation (EME) grant (reference number 14/24/04).

Serious adverse events

Since the start of the trial, a total of 597 SAEs have been reported to the TEAMM trial office (of these, 308 were from patients on levofloxacin and 289 were from patients on placebo). Tables 25–28 present information relating to the type of event by treatment for all SAEs and summarises the severity and causality assessments, and outcomes of each event by treatment.

In summary, 597 SAEs were reported for patients (levofloxacin, n = 308; placebo, n = 289), with the majority reported as being unlikely to be related to or unrelated to the study drug (537/597, 90%) but, instead, related to hospitalisation or prolongation of existing hospitalisation (462/597, 77%). Toxicity reported to be related to levofloxacin (38 episodes) compared with placebo (22 episodes) had a trend towards more gastrointestinal disorders (fours events of nausea/vomiting on levofloxacin vs. one event on placebo) but fewer episodes of diarrhoea (five episodes on levofloxacin vs. nine episodes on placebo), rash (six episodes on levofloxacin vs. one episode on placebo), psychiatric disorders (two episodes on levofloxacin vs. no episodes on placebo) and musculoskeletal and connective tissue disorders (seven episodes on levofloxacin and one episode on placebo) (see Table 44). Appendix 2 provides further information as regards the summary of the CTCAE categories of all SAEs reported (see Table 43) together with their terms of SAEs by treatment (see Table 43). In addition, reported SARs are detailed (see Tables 44 and 45).

Quality of life

Within the scope of the TEAMM grant, QoL was measured at baseline and at 4, 8, 12 and 16 weeks for patients who consented to fill in the QoL booklets. These data include the responses to the EQ-5D, which are reported in Chapter 4, as well as to the EORTC QLQ-C30, EORTC QLQ-MY24 and HADS.

Introduction

An economic evaluation was conducted to estimate the cost-effectiveness of prophylactic levofloxacin compared with placebo in newly diagnosed symptomatic myeloma. The economic evaluation was conducted alongside the TEAMM clinical trial so that only the data collected during the trial were analysed. Cost and outcome data were collected from trial participants for 16 weeks. However, if the intervention is successful in reducing mortality, the benefits may extend beyond this period. Consequently, the economic evaluation includes a within-trial cost-effectiveness analysis over 16 weeks but also explores longer-term outcomes using an analysis extrapolated to 12 months from randomisation.

Methods

Results

Cost-effectiveness results

Cost-effectiveness results are presented in Table 35, which shows the costs and QALYs gained for each treatment arm, the incremental costs and QALYs, and the resulting ICER. The levofloxacin group had the highest QALYs gained over the trial period. The mean total cost was also highest for the levofloxacin group.

TABLE 35 - Cost-effectiveness results

Treatment arm	Cost (£)		QALY		ICER (£/QALY)
	Mean (SD)	Incremental	Mean (SD)	Incremental
Placebo	3272.19 (4173.23)		0.1818 (0.006)
Levofloxacin	3871.45 (4622.02)	599.26	0.1845 (0.006)	0.0026	231,377.42

The results suggest that the use of prophylactic levofloxacin would not be cost-effective compared with placebo. This is as a result of smaller QALY gains and higher costs in the levofloxacin arm than in the placebo arm. An ICER value of £231,377.42 per QALY gain is yielded, which is well above the NICE cost per QALY threshold.

Bootstrapped estimates of the incremental costs and incremental effects are plotted on the cost-effectiveness plane in Figure 7. This shows the joint distribution of incremental costs and incremental effects in the cost-effectiveness plane for levofloxacin compared with placebo. Most of the points in the cloud lie above the cost-effectiveness threshold line, indicating that the use of levofloxacin is unlikely to be a cost-effective use of resources. Although 98.2% of the iterations fall in the north-east quadrant (more costly, more beneficial), only 1.8% fall in the south-east quadrant (less costly, more beneficial).

FIGURE 7.

The cost-effectiveness plane: levofloxacin vs. placebo.

Net monetary benefit

The NMB for each treatment arm, calculated from the bootstrapped estimates of costs and QALYs, is presented in Table 36. Given the decision rule, the NMB results indicate that the use of prophylactic levofloxacin is not a cost-effective use of resources because the expected value of NMB is negative.

TABLE 36 - Net monetary benefit (λ = £20,000)

Treatment arm	Expected value NMB (£)	Standard error	95% CI
Levofloxacin	–184.74	2.10	–188.87 to –180.62
Placebo	361.43	1.89	357.72 to 365.14

The probability that the treatments are cost-effective is presented on the CEAC shown in Figure 8. This shows that the use of prophylactic levofloxacin is very unlikely to be a cost-effective use of resources even at high cost-effectiveness threshold values.

FIGURE 8.

Cost-effectiveness acceptability curve: levofloxacin vs. placebo.

Value-of-information analysis

The population EVPI at the NICE cost-effectiveness threshold value of £20,000 per QALY gained is £99,131. The population EVPI for other values of the cost-effectiveness threshold is plotted in Figure 9.

FIGURE 9.

Expected value of perfect information.

Discussion

Interpretation

The trial indicated that prophylactic levofloxacin caused a significant reduction in febrile episodes and deaths during the 12-week treatment period (log-rank = 9.78; p = 0.002). The largest benefit was observed within 4–8 weeks. Levofloxacin also reduced the number of other (non-febrile) infectious episodes (p-trend = 0.06). The data gathered on these ‘other infections’ are of particular interest because these data have not, to our knowledge, been previously collected on such a large scale. Our data show that these ‘other infections’ are a substantial burden for both patients and the health-care system. They were more frequent than febrile episodes (323 other infections vs. 264 febrile episodes). Again, the benefit of prophylaxis in these ‘other infections’ became apparent during the 4- to 8-week period and continued up to 12 weeks. The body site experiencing the greatest reduction in infections was the urinary tract, probably because levofloxacin is broadly effective against Gram-negative organisms. Upper respiratory tract infections are frequently caused by viruses and, therefore, would be unaffected by levofloxacin.

Information on infections was collected only up to 12 weeks but longer-term survival data were collected up to 12 months. During months 3–6, patients were not on prophylaxis and yet the survival benefit continued for those who had taken levofloxacin. If levofloxacin had prevented death in those patients destined to respond to antimyeloma treatment and subsequently survive beyond 12 months, the survival benefit might have been expected to continue. A possible interpretation of our finding that the survival curves came together by 6 months is that levofloxacin may reduce or delay the risk of death in patients with poorly responsive or refractory disease. Therefore, the cause of death was analysed in the 116 deaths reported during the 12-month trial period. There were fewer deaths attributed to infection in the levofloxacin arm (three patients) than in the placebo arm (eight patients) during the 12-week treatment period, but more in the levofloxacin arm (19 patients) than in the placebo arm (12 patients) post 12 weeks. These data support our possible interpretation outlined above. Although levofloxacin may have reduced the number of deaths in the first 12 weeks from 22 to 8, there were a further 86 deaths in the next 40 weeks from many causes and the reversal of levofloxacin survival advantage may have been influenced by the volume of the later deaths.

There was little benefit from levofloxacin until after the first 4 weeks, and this time delay in benefit from prophylaxis in reducing febrile episodes and other infections is intriguing. The response to treatment of an established infection with antibiotics is expected within 48 hours. This observed delay in benefit from prophylaxis may suggest that some form of biological process needs to take place before the benefit is realised. One suggestion may be that a change in the patient’s immune response or microbiome may be necessary before the benefit of prophylaxis is established. The theory that a biological process may take place in the patient before the benefit of prophylaxis becomes apparent may also be supported by the observation that the survival curves remained separated for months 3–6. It is possible that the benefit of levofloxacin prophylaxis may continue to reduce infections for a period after withdrawing levofloxacin, but this was not measured in the trial.

There was no significant increase in HCAIs in patients on levofloxacin during the 16 weeks of the trial, disproving the theory that the use of prophylactic levofloxacin may trigger an excess of HCAIs. In addition, despite prospectively looking for colonisation with resistant organisms, there were fewer episodes of colonisation noted in the levofloxacin arm. However, it is possible that a longer period of levofloxacin prophylaxis may induce colonisation or infection with HACIs. There is a need to better understand the reasons for the risks of colonisation with resistant organisms and infections in the first 4 weeks of levofloxacin prophylaxis. There is also a need to explore the effects of a longer duration of levofloxacin prophylaxis.

A total of 597 SAEs were reported (308 in the levofloxacin group and 289 in the placebo group), with most reported as unlikely to be related to or unrelated to the study drug (537/597 events, 90%) and a majority reported to be attributable to hospitalisation or prolongation of existing hospitalisation (462/597 events, 77%). Toxicity reported to be related to levofloxacin (38 episodes) compared with placebo (22 episodes) had a tendency to manifest as gastrointestinal disorders, rash, psychiatric disorders and musculoskeletal/connective tissue disorders, but these were very rare events.

The health economic analysis showed a slight increase in QALYs over 16 weeks for levofloxacin prophylaxis, but this was associated with higher health-care costs. Despite the reduction in febrile episodes and deaths in the patients taking prophylactic levofloxacin, the 16-week cost-effectiveness analysis indicated that prophylactic levofloxacin would not be a cost-effective use of NHS resources. This result was driven by higher use of health-care resources in the levofloxacin arm and only very small gains in QALYs. Consequently, the aforementioned reduction in febrile episodes and deaths did not translate into gains in QALYs that were sufficient to outweigh the additional costs. This result was robust to sensitivity analyses. As the benefit of levofloxacin prophylaxis in reducing infections and death became most apparent after 4 weeks, it would be useful to perform the analysis over the period of 4–16 weeks. This analysis is planned for the future. In addition, the analysis of 12-month data is ongoing. More research is needed to investigate whether or not the cost for avoided febrile episode is cost-effective. 56

Generalisability

Overall evidence

This study supports the use of levofloxacin as prophylaxis for myeloma patients undergoing antimyeloma treatment. A small study of prophylactic co-trimoxazole in early myeloma showed a reduction in bacterial infections but was too small to detect reduced mortality and 25% of patients were unable to tolerate the drug. 18 Two large studies22–24,58 confirmed the efficacy of prophylactic levofloxacin in preventing bacterial infection in neutropenia, and subsequent meta-analyses showed superiority of fluoroquinolone antibiotics over other classes of antibiotic.

Overall conclusions

In the authors’ opinion, this study supports the use of prophylactic levofloxacin for patients undergoing antimyeloma treatment. This is the largest study in the literature assessing the use of levofloxacin to reduce febrile episodes and other infections for myeloma patients. Most patients in the trial were fitter then expected with a lower carriage rate of C. difficile.

There was no difference for new acquisitions of or invasive infections by C. difficile, MRSA and ESBL Gram-negative organisms when assessed up to 16 weeks.

Despite being clinically effective, levofloxacin in prophylaxis is not a cost-effective health-care programme from the NHS standpoint. However, this result is focused only on incremental QALYs gained, which are usually negligible when measured for such a short period of time (16 weeks). More research is needed to investigate whether or not the cost for avoided febrile episode is cost-effective.

Further research

Laboratory investigation of immunity, inflammation and disease activity on stored samples (funded by the TEAMM NIHR EME programme grant reference number 14/24/04) will provide insights into the triad of (1) myeloma disease activity, (2) immune competence and (3) infections and HCAIs to stratify patients for risk of infection, to guide use of prophylactic antibiotics and to improve response to antimyeloma therapies.

Further research to establish the optimal duration of prophylactic antibiotics for these patients is needed. There may be a benefit to treat for the whole duration of antimyeloma treatment in order to give maximum protection for these patients. In addition, the use of Septrin gave additional benefit to levofloxicin. This needs to be investigated within a future clinical trial.

Co-applicants

Mark T Drayson, Stella Bowcock, Janet A Dunn, Tim Planche, Gulnaz Iqbal, Guy Pratt, Kwee Yong, Eric Low, Peter Hawkey, Doug Carroll (QoL advisor) and Claire T Hulme.

Trial Management Group

Mark T Drayson, Stella Bowcock, Tim Planche, Kwee Yong, Guy Pratt, Eric Low, Kerry Raynes, Gulnaz Iqbal, Janet A Dunn, David Meads, Bryony Dawkins, Claire T Hulme and Helen Higgins.

Warwick Clinical Trials Unit staff working on TEAMM trial

Janet A Dunn, Gulnaz Iqbal, Jill Wood, Helen Higgins, Kerry Raynes, Joanne O’Beirne-Elliman, Uzma Manazar, Lauren Betteley, Pankaj Mistry, Catherine Hill, Kiran Bal, David Boss, Bushra Rahman, Sarah Lowe, Kimberly White, Penelope Goode, Claire Daffern, Amy Ismay and Joanne Grummet.

Independent review of early deaths

Graham Jackson provided an independent review of deaths within the first 12 weeks.

Drug supply

Oliver Gupta [Modepharma Limited (Beckenham, UK)] organised the manufacturing and supply of levofloxacin and placebo tablets.

Data Monitoring Committee members

Anthony Child, Tim Boswell and Walter Gregory.

Independent Trial Steering Committee members

Mark Wilcox, Roger Owen and Steven Devereux.

Microbiology

Tim Planche and Irene Monahan at St George’s Hospital, University of London.

Central laboratory Birmingham

Mark T Drayson, Tim Plant, Karen Walker, Nicki Newnham, Alison Adkins, Jennifer Heaney and Zaheer Afzal.

Health economists

David Meads, Bryony Dawkins and Claire T Hulme.

Patient and public involvement

Eric Low was a co-applicant, member of the Trial Management Committee and Trial Steering Committee and, together with colleagues from Myeloma UK, commented on the drafting of the protocol and PIS as well as the main report. We would also like to acknowledge all of the 977 trial participants, especially those who returned their patient diaries and QoL booklets.

The local NHS teams

The TEAMM team would like to thank the NHS trusts and participating PIs and their colleagues for recruiting and monitoring trial participants. These include [in alphabetical order of participating hospitals preceded by the local PI(s)] Feargal McNicholl of Altnagelvin Hospital; Philip Windrum of Antrim Hospital; Joanne Howard and Parag Jasani of Basildon University Hospital; Sylwia Simpson of Basingstoke and North Hampshire Hospital; Paul Kettle and Oonagh Sheehy of Belfast City Hospital; Adrian Williams of Bradford Royal Infirmary; Vijoy Chowdhury of Broomfield Hospital; Sylvia Feyler, Kate Rothwell and Srivasavi Dukka of Calderdale Royal Hospital; Haz Sayala and David Allsup of Castle Hill Hospital; Sally Chown of Cheltenham General Hospital; Robert Cutting, Emma Welch and Peter Toth of Chesterfield Royal Hospital; Michael Hamblin and Gavin Campbell of Colchester General Hospital; Salaheddin Tueger of Countess of Chester Hospital; Aurangzeb Razzak and Kamaraj Karunanithi of County Hospital stafford; Kathryn Boyd of Craigavon Area Hospital; Anil Kamat and Tariq Shafi of Darent Valley Hospital; John Ashcroft of Dewsbury Hospital; Susan Levison-Keating of Diana, Princess of Wales Hospital; Akeel Moosa of Dorset County Hospital; Richard Kaczmarski and Taku Sugai of Ealing Hospital; Alistair Whiteway of Frenchay Hospital; Jagadeesh Gandla and Jhansi Muddana of George Eliot Hospital; Earnest Heartin of Glan Clwyd Hospital; Sally Chown of Gloucestershire Royal Hospital; Bhuvan Kishore of Good Hope Hospital; Sandor Lueff, Kandeepan Saravanamuttu, Caroline Harvey and Charlotte Kallmeyer of Grantham Hospital; Norbert Blesing of Great Western Hospital; Matthew Streetly of Guy’s and St Thomas’ Hospital; Guy Pratt and Bhuvan Kishore of Heartlands Hospital; Lisa Robinson of Hereford County Hospital; Richard Kaczmarski of Hillingdon Hospital; Sylvia Feyler, Kate Rothwell and Srivasavi Dukka of Huddersfield Royal Infirmary; Mark Kwan of Kettering General Hospital; Steve Schey and Stella Bowcock of King’s College Hospital; Rowena Faulkner and Tim Moorby of King’s Mill Hospital; Samir Zebari of Kingston Hospital; Claire Chapman and Mamta Garg of Leicester Royal Infirmary; Kamaraj Karunanithi of Leighton Hospital; Sandor Lueff, Kandeepan Saravanamuttu, Caroline Harvey and Charlotte Kallmeyer of Lincoln County Hospital; John Hudson of Macclesfield District General Hospital; Simon Gibbs, Eleni Tholouli and Alberto Rocci of Manchester Royal Infirmary; Vivienne Andrews, Vijayavalli Dhanapal and Sarah Arnott of Medway Maritime Hospital; Moez Dungarwalla of Milton Keynes Hospital; Nilima Parry-Jones of Nevill Hall Hospital; Abraham Jacobs and Supratik Basu of New Cross Hospital; Claudius Rudin of North Devon District Hospital; Neil Rabin of North Middlesex University Hospital; Jane Parker of Northampton General Hospital; Charalampia Kyriakou of Northwick Park Hospital; Sandor Lueff, Kandeepan Saravanamuttu, Caroline Harvey and Charlotte Kallmeyer of Pilgrim Hospital; John Ashcroft of Pinderfields Hospital; John Ashcroft of Pontefract Hospital; Rebecca Maddams and Fergus Jack of Poole Hospital; Stella Bowcock of Princess Royal University Hospital; Helen Dignum and Edward Belsham of Queen Alexandra Hospital; Mark Cook of Queen Elizabeth Hospital; Ramaprabhahari Sathesh-Kumar and Martin Lewis of Queen Elizabeth Hospital, King’s Lynn; Tullie Yeghen of Queen Elizabeth Hospital, Woolwich; Humayun Ahmad of Queen’s Hospital, Burton upon Trent; Biju Krishnan and Sandra Hassan of Queen’s Hospital, Romford; Karthick Ramasamy and Henri Grech of Royal Berkshire; Rachel Hall of Royal Bournemouth Hospital; Claudius Rudin of Royal Devon & Exeter Hospital; Helen Jackson of Royal Gwent Hospital; Yousef Ezaydi of Royal Hallamshire Hospital; Katharine Lowndes of Royal Hampshire County Hospital; Stephen Hawkins of Royal Liverpool University Hospital; Kamaraj Karunanithi of Royal Stoke University Hospital; Louise Hendry of Royal Surrey County Hospital; Sally Moore of Royal United Hospital Bath; Jeff Neilson of Russells Hall Hospital; Jonathan Cullis of Salisbury District Hospital; Farooq Wandroo of Sandwell General Hospital; Matthew Jenner of Southampton General Hospital; Paul Cervi of Southend University Hospital; Alastair Whiteway of Southmead Hospital; Simon Stern of St Helier Hospital; Gordon Cook of St James University Hospital; Robin Aitchison of Stoke Mandeville Hospital; Rangaiah Madhusadan and Shikha Chattree of Sunderland Royal Hospital; Hussen Baden of Tameside General Hospital; Andreea Corcoz of The Royal Shrewsbury Hospital; Deborah Turner and Heather Eve of Torbay Hospital; Moulod El-Agnaf of Ulster Hospital; Anand Lokare and Beth Harrison of University Hospital Coventry; Tullie Yeghen of University Hospital Lewisham; Simon Watt of University Hospital South Manchester; Chandramouli Nagarajan, Mohamed Kaleel-Rahman and Jayaprakash Ramachandran of Warrington Hospital; Anton Borg and Caroline Arbuthnot of Warwick Hospital; Magda Jabbar Al-Obaidi of West Middlesex University Hospital; Peter Cumber of West Wales General Hospital; Mark Offer of Wexham Park Hospital; Naim Akhtar of Whipps Cross Hospital; Sumant Kundu of Withybush Hospital; Lally Desoya of Wrexham Maleor Hospital; and Robin Aitchison of Wycombe Hospital.

Other acknowledgements

The TEAMM team would like to thank the NIHR programme for funding this research.

The project benefited from facilities funded through Advantage West Midlands Science City Programme. The project received support through the Clinical Research Networks.

Contributions of authors

Mark T Drayson (Chief Investigator) was responsible for identifying the research question and writing the trial protocol, the central laboratory analysis and the review of the blood and urine samples.

Stella Bowcock (Co-Chief Investigator) was responsible for identifying the research question and writing the trial protocol.

Tim Planche (Microbiology Advisor) was responsible for identifying the research question and writing the trial protocol, provided expert microbiology advice and performed the analysis on the stool samples and nasal swabs.

Gulnaz Iqbal (Trial Statistician) was responsible for identifying the research question and writing the trial protocol, and carried out the statistical analysis and interpreted the results.

Guy Pratt (Haematologist) was responsible for identifying the research question and writing the trial protocol.

Kwee Yong (Haematologist) was responsible for identifying the research question and writing the trial protocol.

Jill Wood (Trial Co-ordinator) managed the co-ordination and administration of the trial.

Kerry Raynes (Trial Co-ordinator) managed the co-ordination and administration of the trial.

Helen Higgins (Senior Project Manager) managed the co-ordination and administration of the trial.

Bryony Dawkins (Health Economics) conducted the health economic analysis.

David Meads (Health Economics) conducted the health economic analysis.

Claire T Hulme (Health Economics) oversaw the health economic analysis.

Anna C Whittaker (QoL Advisor) provided expert advice on QoL.

Peter Hawkey (Microbiology) performed additional microbiology tests.

Eric Low (Patient Advocate, Myeloma UK) was responsible for all the patient and public involvement activities throughout the trials.

Janet A Dunn (WCTU Lead) was responsible for identifying the research question and writing the trial protocol, and carried out the statistical analysis and interpreted the results.

All authors provided comments on the report and approved the final version.

Publications

Dunn JA, Harris B, Iqbal G, Wood J, Mistry P, O’Beirne-Elliman J, et al. Use of Patient Diaries in Conjunction with Standard Reporting Methods: Duplication of Data or a Valuable Resource? Paper presented at the National Cancer Research Institute conference, Liverpool, November 2015. Abstract no. 141.

Dunn JA, Harris B, Iqbal G, Wood J, Mistry P, O’Beirne-Elliman J, et al. Use of Patient Diaries in Conjunction with Standard Reporting Methods: Duplication of Data or a Valuable Resource? Paper presented at the Third International Clinical Trials and Methodology conference, Glasgow, 16–17 November 2015. Abstract no. 183.

Bowcock S, Atkin C, Iqbal G, Planche T, Pratt G, Yong K, et al. Diagnostic Pathways of Myeloma Patients Presenting to Hospital Care and Relationship to end organ Damage: An Analysis from the TEAMM (Tackling EArly Morbidity and Mortality in Myeloma) Trial in 977 Patients. Paper presented at the American Society of Haematology conference, Atlanta, GA, December 2017. Abstract no. 2132.

Chicca IJ, Heaney JLJ, Iqbal G, Bowcock S, Planche T, Wood J, et al. Assessment of anti-Bacterial Antibodies in Multiple Myeloma Patients at Disease Presentation and in Response to Therapy: Implication for Patient Management. Paper presented at the American Society of Haematology conference, Atlanta, GA, December 2017. Abstract no. 3036.

Drayson MT, Bowcock S, Planche T, Iqbal G, Wood J, Raynes K, et al. Tackling EArly Morbidity and Mortality in Myeloma: Assessing the Benefit of Antibiotic Prophylaxis and its Effect on Healthcare Associated Infections. Paper presented at the American Society of Haematology conference, Atlanta, GA, December 2017. Abstract no. 903.

Atkin C, Drayson MT, Iqbal G, Yong K, Pratt G, Planche T, et al. Presenting Symptoms of Patients with Multiple Myeloma: Data from the TEAMM (Tackling EArly Morbidity and Mortality in Myeloma) Trial. Paper presented at the British Society of Haematology conference, Liverpool, April 2018. Abstract no. 243.

Bowcock S, Atkin C, Iqbal G, Yong K, Pratt G, Planche T, et al. Does a Long Diagnostic Pathway in Myeloma Lead to More Advanced Disease At Presentation? An Analysis from the TEAMM trial (Tackling EArly Morbidity and Mortality in Myeloma). Paper presented at the British Society of Haematology conference, Liverpool, April 2018. Abstract no. 428.

Bowcock S, Planche T, Iqbal G, Pratt G, Yong K, Wood J, et al. Levofloxacin Prophylaxis in Newly Diagnosed Myeloma Reduces Febrile Episodes and Death without Increasing Healthcare Associated Infections: Results from the TEAMM Trial. Paper presented at the 23rd Congress of the European Hematology Association, Stockholm, June 2018. Abstract no. 2453.

Bowcock S, Planche T, Iqbal G, Pratt G, Yong K, Wood J, et al. Levofloxacin Prophylaxis in Newly Diagnosed Myeloma Reduces Febrile Episodes and Death Without Increasing Healthcare Associated Infections: Results From the TEAMM Trial (Tackling EArly Morbidity and Mortality in Myeloma). Paper presented at the British Society of Haematology conference, Liverpool, April 2018. Abstract no. 234.

Helwick C. For Patients Treated for Myeloma, Antibiotic Prophylaxis May Reduce Infections and Deaths. The ASCO Post. 10 February 2018. URL: www.ascopost.com/issues/february-10-2018/for-patients-treated-for-myeloma-antibiotic-prophylaxis-may-reduce-infections-and-deaths/ (accessed 2 August 2018).

Planche T, Iqbal G, Bowcock S, Wood J, Raynes K, Monahan I, et al. Levofloxacin Prophylaxis – Assessing the Effects and Microbiology of Invasive Infections in a Phase III Study of 977 Patients with Newly Diagnosed Multiple Myeloma. Paper presented at the 28th Annual Congress of the European Congress Of Clinical Microbiology And Infectious Diseases, Madrid, April 2018. Abstract no. 4447.

Planche T, Iqbal G, Bowcock S, Wood J, Raynes K, Monahan I, et al. Levofloxain Prophylaxis does not Increase the Carriage of Resistant Organisms – Tackling EArly Mobidity and Mortality in Myeloma (TEAMM) Phase III Study: Assessing the Benefits of Antibiotic Prophylaxis and its Effect on Carriage of Resistant Organisms in 977 Patients. Paper presented at the 28th Annual Congress of the European Congress Of Clinical Microbiology And Infectious Diseases, Madrid, April 2018. Abstract no. 4479.

Drayson MT, Bowcock S, Planche T, Iqbal G, Pratt G, Yong K, et al. Levofloxacin prophylaxis in patients with newly diagnosed myeloma (TEAMM): a multicentre, double-blind, placebo-controlled, randomised, phase 3 trial [published online ahead of print October 23 2019]. Lancet Oncol 2019.

Ramasamy K. Reducing infection-related morbidity and mortality in patients with myeloma [comment] [published online ahead of print October 23 2019]. Lancet Oncol 2019.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.